I think the idea is that almost all of the costs like management, lobbying, paywalls, etc. are to be done away with, which that article estimates is at least 50

I think the idea is that almost all of the costs like management, lobbying, paywalls, etc. are to be done away with, which that article estimates is at least 50% of the cost. Getting rid of a profit margin by being a public non-profit org slashes another 30%, and so presumably a system like this may only cost a few hundred dollars an article on average, well below the thousands of dollars typically charged for top journals. At that point it's easy for costs to be handled by libraries and institutions contributing to operating costs at much less than (the author of the OP claims <1% but I'd guess more like ~20%) the cost of existing journal access and publishing fees paid by institutions today.